//01  Requesty alternative

At or below list,
on every token.

Requesty bills each model at its provider’s rate plus 5%, or 0% with your own provider keys. On SayGM, providers compete to serve your requests, so Claude, GPT and Gemini bill at or below the model maker’s own list price, through a gateway sealed inside a Trusted Execution Environment.

Maker's list priceper 1M tokens$10.00
Through Requestylist plus their 5% markup$10.50
Through SayGMat or below list≤ $10.00
Example from Requesty’s pricing page
Typical saving · 9 models19%below the same model on Requesty, with their 5% markup counted.
AvailableNative OpenAI, Anthropic and Gemini APIs
[02] The price

At or below list price,
by design.

SayGM prices every model at or below the model maker’s published list price, at any volume and from your first request. Independent providers compete to serve each request, and that competition sets the rate.

Funding your balance costs the amount you fund it with. You pay the rate of the provider that serves each request, never above list.

Requesty• Provider price + 5% on model cost0% with your own provider keys.
SayGMAt or below list · top-ups credited in fullCapped at the model maker’s list price.
[03] The prompt

“Where does this data go?” deserves an answer you can verify.

Requesty gives you data controls: prompt logging can be switched off per key, and an EU gateway in Frankfurt keeps their own processing in the EU. SayGM works at a different layer. Its gateway runs inside an Intel TDX trusted execution environment, and the attestation quote proving it is independently verifiable, so its operators and host machines cannot read the API requests passing through it.

Claude, GPT and Gemini process your prompt under their own API terms, exactly as they would if you called them directly. For inference sealed end to end, SayGM’s confidential open models run inside TEEs from gateway to model.

Verifiable by anyone

Remote attestation lets you confirm for yourself exactly what code the gateway is running.

Sealed from the routing layer too

Most privacy claims stop at the model. This one covers the gateway itself.

Built for the audit trail

Pairs with signed logs of which model handled what, for teams that need to show their work.
[04] Calculator

What you could save.

Pick a model, set your monthly volume, see the delta.
Model
50M
15M
91%

Most input tokens in real traffic are cache reads, and both gateways price them well below a fresh one — so the share matters more to the bill than the totals do. The default is what SayGM measures across its own traffic.

SayGM$367.66
Requesty$442.21
You keep$74.54per month · 16% lower$894.52 over a yearstart saving
[05] Model by model

Price per million tokens.

USD per 1M tokens, uncached input / output. SayGM’s side is the best rate a provider offered, as of 59 minutes ago.Requesty’s is the rate their public model list publishes, plus their 5% markup, as of 33 minutes ago.
ModelSayGMRequestyYou save
Claude Opus 5Anthropic$4.37 / $21.83$5.25 / $26.2516%
Claude Sonnet 5Anthropic$1.75 / $8.73$2.10 / $10.5016%
Claude Haiku 4.5Anthropic$0.87 / $4.37$1.05 / $5.2516%
GPT-5.5OpenAI$4.08 / $24.45$5.25 / $31.5022%
GPT-5.6 TerraOpenAI$1.63 / $9.78$2.10 / $12.6022%
GPT-5.6 LunaOpenAI$0.16 / $0.98$0.21 / $1.2622%
Gemini 3.1 Pro PreviewGoogle$1.70 / $10.20$2.10 / $12.6019%
Gemini 3.5 FlashGoogle$1.28 / $7.65$1.58 / $9.4519%
GLM-5.3Z.ai$0.80 / $2.53$1.47 / $4.6245%
browse the full catalogue
[06] Side by side

SayGM vs. Requesty, line by line.

DimensionSayGMvs. Requesty
PricingAt or below the model maker's list price, top-ups credited in fullProvider price plus 5% on usage, or 0% with your own provider keys
Gateway privacyAPI requests pass through a gateway sealed in an Intel TDX TEEEncrypted in transit and at rest; prompt logging on by default for 30 days, off per key, with org-wide zero retention on written request
ProofA remote-attestation quote anyone can verifyPublished security page, DPA on request, SOC 2 Type II programme in progress
Model coverageClaude, GPT, Gemini and a curated set of open models600+ models across 20+ providers, by their count
Routing controlsBuild cascade models that try the next model when one fails, or fusion models that combine several. Every request is balanced across providers by latency, reliability and price, retried on another provider if an attempt fails, and kept on one provider through a conversation where it can, so prompt caching holdsConfigurable fallback chains, weighted load balancing and latency routing
AnalyticsUsage and spend in the dashboardReal-time cost and latency analytics, grouped by model, user, provider or header
Enterprise governanceSelf-serve accounts and API keysSSO, RBAC and custom SLAs on the Enterprise plan
Data residencySealed gateway; inference runs where each provider serves itEU gateway in Frankfurt on every plan, plus EU-region models
Free tierPrepaid, pay as you goFree models at 200 requests a day
API surfaceEach model on its native API: OpenAI, Anthropic Messages and GeminiOpenAI-compatible for every model, plus Anthropic Messages and Responses
Requesty details as of September 2026, from their own pricing, security and documentation pages. SayGM lists 56 models today. If the one you need is not among them, or your team needs SSO on day one, Requesty is the better tool for that job, and a second route to the same models is worth having whichever gateway you make primary.
[07] Questions

Common questions about switching.

How does Requesty price models?

As of September 2026, their pricing page says they add 5% to the base model provider cost: a model that costs $10 per 1M tokens costs $10.50 through them. Their free tier covers free models at 200 requests a day. The details are on their pricing page. SayGM bills each model at or below its maker's list price, and every top-up is credited in full.

How much cheaper is SayGM than Requesty?

Across the models this page compares, 19% for the middle one, counting Requesty's 5% markup. It varies by model, and the calculator above prices your exact volumes. Prices differ across SayGM's wider catalogue, so check the model you run.

I use Requesty with my own provider keys. Does SayGM still save me money?

With your own keys, Requesty's gateway charge is 0%, according to their enterprise page (as of September 2026), and you pay your provider directly at your own account's rate. SayGM bills at or below the published list price, so compare it with that rate: a negotiated discount or a provider free tier can come in lower.

Does Requesty keep my prompts?

As of September 2026, their security page says prompt and output logging is on by default on self-serve plans, encrypted and kept in the EU for up to 30 days. You can turn it off per API key, and organisation-wide zero retention is available on written request. See their security page. SayGM's API requests pass through a gateway sealed inside an Intel TDX trusted execution environment, so its operators and host machines cannot read them. Claude, GPT and Gemini then process your prompt under their own API terms, as they would if you called them directly.

Which has more models?

Requesty, by a wide margin. Their pricing page lists 600+ models across 20+ providers as of September 2026. SayGM carries a curated catalogue: Claude, GPT and Gemini plus a set of open models, several of them served inside TEEs end to end.

I need SSO and EU data residency. Which should I pick?

Requesty offers both today. SSO and RBAC come with their Enterprise plan, and EU data residency through a Frankfurt gateway is on every plan; their EU routing docs explain that keeping inference in the EU also means choosing EU-region models. SayGM is self-serve, and its privacy guarantee comes from attested hardware rather than from a region. If a procurement checklist asks for SSO and EU residency now, Requesty meets it.

Can I control routing the way I do on Requesty?

In a different way. Requesty lets you set up fallback chains, weighted load balancing and latency routing yourself, as their routing policy docs describe. SayGM routes every request automatically: it balances each one across providers by latency, reliability and price, retries a failed attempt on another provider, and keeps a conversation on one provider where it can, so prompt caching holds. On top of that you can build your own models under My Models in the dashboard: a cascade model tries the next model in your list when one fails, and a fusion model asks several models and combines their answers. Each gets one model id you call like any other. If you want to hand-tune provider weights, Requesty gives you that control.

How do I switch from Requesty?

Point your SDK's base URL at the SayGM gateway and use a SayGM API key. Claude runs on the Anthropic Messages API, GPT on the OpenAI API and Gemini on its native API, the same shapes as the SDKs you already use. If you call Claude or Gemini through an OpenAI-compatible client on Requesty today, switch those calls to the Anthropic or Gemini SDK. Use the model ids listed in SayGM's catalogue.